Abstract:
Wavelet support vector regression is utilized to enhance the resolution for remote sensing multispectral image. Firstly, both low resolution multispectral image and high resolution panchromatic image are decomposed into multiresolution by using nonsubsampled contourlet transform. Then, by using wavelet support vector regression, the decomposed coefficients are learned and predicted so as to obtain multispectral image with preliminary enhanced resolution. Finally, the above results are further fused with the traditional interpolate one to achieve the resolution enhance of multispectral image. Experiment results show that the proposed algorithm utilizes the auxiliary of remote sensing panchromatic image to effectively attain a wealth of highfrequency detail information, such that either the minimum mean squared error or the peak signal to noise ratio is superior to these from the traditional methods only depending on the amplification of image itself and other resolution enhance methods.